[Breaking] AWS App Runner Ends New Customer Access on April 30. Migrate to ECS Express Mode
AWS will end new customer access to App Runner on April 30, 2026. Existing users can continue but no new features will be added.
News
kkm
Backend Engineer / AWS / Django
AWS will end new customer access to App Runner on April 30, 2026. Existing users can continue but no new features will be added.
AWS will end new customer access to App Runner on April 30, 2026. Existing users can continue using the service, but no new features will be added. The recommended migration path is Amazon ECS Express Mode, announced at re:Invent 2025 in November 2025.
App Runner launched in May 2021 as the simplest way to deploy containerized applications on AWS. Its appeal was straightforward: hand over a container image or source code, and your app goes live without configuring load balancers or auto-scaling.
What Happened
According to AWS official documentation, after "careful consideration," AWS decided to end new customer access to App Runner and "focus on delivering the innovations where customers find the most value."
| Item | Details |
|---|---|
| Service | AWS App Runner |
| New Customer Cutoff | April 30, 2026 |
| Existing Users | Can continue using the service (including creating new resources) |
| New Features | None (security and availability investments continue) |
| Recommended Migration | Amazon ECS Express Mode |
No complete shutdown date has been announced. This is specifically about ending new customer sign-ups — existing App Runner users can keep running their services after April 30. However, new AWS accounts will no longer be able to start using App Runner.
End of New Sign-ups ≠ Immediate Shutdown. So When Will It Actually Stop?
The first question on everyone\'s mind is: "So when will it actually become unusable?" The short answer: no complete shutdown date has been announced yet.
According to the AWS Service Lifecycle, services are retired in three phases.
| Phase | Description | App Runner Status |
|---|---|---|
| ① Maintenance | New customer access ends. Existing users continue normally. No new features | ← Current phase |
| ② Sunset | Shutdown announced. Typically 12 months notice | Not reached |
| ③ Full Shutdown | Service completely removed. No longer available in any capacity | Not reached |
App Runner has just entered the Maintenance phase (①). Before it can reach Full Shutdown (③), a Sunset announcement (②) is required, which typically comes with 12 months of lead time. Even if Sunset were announced tomorrow, you\'d still have at least a year.
So what actually happened to other AWS services that went through the "no new customers" phase?
| Service | New Customer Cutoff | Full Shutdown | Time to Shutdown |
|---|---|---|---|
| OpsWorks Stacks | May 2023 | May 26, 2024 | ~12 months |
| OpsWorks for Chef | ~Feb 2024 | May 5, 2024 | ~3 months |
| OpsWorks for Puppet | ~Feb 2024 | March 31, 2024 | ~2 months |
| CodeStar | July 2024 | Project features ended (resources persist) | — |
| CodeCommit | July 2024 | TBD (still running as of April 2026) | 21+ months and counting |
| Cloud9 | July 2024 | TBD (still running as of April 2026) | 21+ months and counting |
| SimpleDB | Early 2010s | TBD (still running after 10+ years) | 10+ years |
Some services like OpsWorks were fully shut down within 12 months, while others like CodeCommit and SimpleDB have continued running for years in "no new customers" limbo. It\'s impossible to predict which pattern App Runner will follow, but at minimum, a Sunset announcement plus 12 months of grace period is required before the service goes away entirely.
What Should Existing Users Do
First, App Runner won\'t suddenly stop working after April 30. AWS has explicitly stated they will "continue investing in the security and availability" of the service. There\'s no need to rush a migration today.
That said, with no new features being added, App Runner will gradually fall behind competitors like Google Cloud Run and Azure Container Apps. It\'s time to start planning a migration.
AWS offers two main migration paths. The first and recommended option is ECS Express Mode, which inherits App Runner\'s ease of deployment while adding ECS\'s flexibility. AWS\'s official migration guide recommends gradual traffic migration using Route 53 weighted routing (10% → 25% → 50% → 75% → 100%).
The second option is migrating to a standard ECS configuration using AWS CDK L3 constructs or CloudFormation, suitable for teams that need fine-grained infrastructure control.
What Is ECS Express Mode
ECS Express Mode is a new ECS capability announced at re:Invent 2025 in November 2025. Provide a container image and IAM roles, and it automatically sets up an ALB, auto-scaling, SSL/TLS termination, and CloudWatch monitoring.
While similar in concept to App Runner, there\'s a key difference. App Runner completely hid the infrastructure, whereas ECS Express Mode lets users access and modify all auto-provisioned resources (ECS clusters, task definitions, ALBs, security groups, etc.). The design philosophy is "start simple, customize when needed."
| Feature | App Runner | ECS Express Mode |
|---|---|---|
| Infrastructure Visibility | ALB etc. hidden | Full access to all resources |
| Source Code Direct Deploy | Supported (GitHub integration) | Not supported (CI/CD setup required) |
| Scale to Zero | Supported (low cost when idle) | Not supported (minimum 1 task required) |
| Pricing Model | Usage-based | Capacity-based (ALB shared across up to 25 services) |
| Sidecar Containers | Not supported | Supported |
| IaC Support | CloudFormation | CloudFormation, CDK, Terraform |
| Deploy Strategy | Rolling | Canary deployment (default) |
There are important caveats. Two of App Runner\'s biggest draws — source code direct deploy and scale to zero — are not available in ECS Express Mode. Replicating the auto-deploy-on-push experience requires setting up GitHub Actions and writing a Dockerfile separately. Running costs also can\'t be reduced to zero during idle periods due to ECS Express Mode\'s architecture.
How Will Costs Change After Migration?
Cost is unavoidable when evaluating a migration. Here\'s a side-by-side comparison of App Runner and AWS Fargate (the compute layer behind ECS Express Mode) pricing in the Tokyo region.
| Cost Item | App Runner | Fargate (x86) | Fargate (ARM) |
|---|---|---|---|
| vCPU (per hour) | $0.0809 | $0.05056 | $0.04045 |
| Memory (per GB/hour) | $0.00885 | $0.00553 | $0.00442 |
| Scale to Zero | Supported (memory-only billing when idle) | Not supported (minimum 1 task always billed) | |
| Load Balancer (ALB) | Included in price | Separate charge (~$18–25/month + usage fees) | |
On a per-unit basis, Fargate is roughly 37% cheaper for both vCPU and memory compared to App Runner (x86 comparison). But actual costs depend heavily on your workload characteristics.
Here\'s an estimate for a small web app with 1 vCPU and 2 GB memory.
| Scenario | App Runner | ECS Express Mode (Fargate x86) |
|---|---|---|
| Always running (24h × 30 days) | vCPU: $58.25 Memory: $12.74 ALB: $0 Total: ~$71/month | vCPU: $36.40 Memory: $7.96 ALB: ~$20 Total: ~$64/month |
| 8h active per day + 16h idle | Active: $21.50 Idle (memory only): $8.50 ALB: $0 Total: ~$30/month | (No scale-to-zero, same as always running) Total: ~$64/month |
For always-on, high-traffic apps, ECS Express Mode offers clear cost advantages. For apps with long idle periods, App Runner\'s scale-to-zero makes it significantly cheaper. Migration could mean "lower unit prices but higher total costs" — so running your own estimates is essential.
For detailed pricing, see the official pages:
AWS Copilot CLI Also Ending Support in June
Related to App Runner, AWS Copilot CLI will end support on June 12, 2026. Copilot CLI was a tool that simplified building, releasing, and operating container applications on ECS and App Runner.
AWS cited the learning curve of its proprietary CLI and manifest syntax, limited infrastructure visibility, and customization constraints as reasons for ending support. The code will remain available as an open-source GitHub project, but AWS will no longer provide security updates or new features.
App Runner, Copilot CLI, and AWS Proton (full support ending October 2026) — the pattern is clear. AWS\'s "simplified container deployment" services are being consolidated into ECS (+Express Mode) and CDK.
AWS\'s Ongoing Service Consolidation Since 2024
App Runner\'s end of new customer access is part of AWS\'s broader service portfolio consolidation that began in 2024. In July 2024, several services including CodeCommit, Cloud9, and CloudSearch ended new customer access.
| Service | New Customer Cutoff | Recommended Migration |
|---|---|---|
| CodeCommit | July 2024 | GitHub, GitLab |
| Cloud9 | July 2024 | IDE Toolkits, CloudShell |
| CloudSearch | July 2024 | OpenSearch |
| AWS Proton | October 2025 | CloudFormation, CDK |
| App Runner | April 2026 | ECS Express Mode |
| Copilot CLI | June 2026 (EOL) | ECS Express Mode, CDK |
AWS has over 200 services, and overlapping functionality has persisted for years. For simplified container deployment alone, there were App Runner, Copilot CLI, Elastic Beanstalk, Proton, and Lightsail Containers at one point. On GitHub\'s App Runner roadmap, the developer community had noted that feature development nearly stopped in late 2024, making today\'s announcement somewhat predictable.
Developer Community Reactions
Developer reactions to App Runner\'s end are largely unsurprised. On the GitHub roadmap issue, developers had been voicing frustration since early 2025: "2024 saw only 2 feature improvements, down from 17 in 2023" and "HTTP/2 support has been in the backlog so long that HTTP/3 is now being requested."
Reactions to ECS Express Mode are mixed. While many welcome "App Runner-level simplicity for ECS," Hacker News discussions point out that "without scale to zero, it\'s not a true App Runner replacement" and "Cloudflare Containers offers more rational usage-based pricing."
When ECS Express Mode was announced in November 2025, many developers had already predicted App Runner\'s demise.
Summary
App Runner served as "the simplest way to deploy containers" for five years, but is now ceding that role to ECS Express Mode as part of AWS\'s service consolidation.
Existing users don\'t need to panic. Based on AWS\'s service lifecycle, full shutdown requires a separate Sunset announcement plus 12 months of notice — and precedents like CodeCommit show that services can continue running for nearly two years after the new-customer cutoff. However, with no new features coming, migration planning should begin sooner rather than later.
ECS Express Mode aims to balance "simplicity and flexibility," but App Runner\'s unique strengths — source code direct deploy and scale to zero — are not available. On the cost side, always-on workloads may see savings, while apps with long idle periods could actually see higher total costs. Check out the detailed overviews on DevelopersIO and Publickey to evaluate whether it fits your use case.
AWS\'s service consolidation will likely continue. Services that are "convenient but overlap with others" could become future consolidation targets. The best preparation is maintaining portable, container-image-based architectures that aren\'t tightly coupled to any single deployment service.
Sources
- ▸ AWS App Runner availability change - AWS Official Documentation
- ▸ AWS Service Lifecycle - AWS General Reference
- ▸ Announcing Amazon ECS Express Mode - AWS What\'s New (November 21, 2025)
- ▸ Build production-ready applications using Amazon ECS Express Mode - AWS Blog
- ▸ AWS App Runner Pricing
- ▸ AWS Fargate Pricing
- ▸ Announcing the end of support for the AWS Copilot CLI - AWS Containers Blog
- ▸ App Runner Roadmap Issue #274 - GitHub
- ▸ ECS Express Mode - DevelopersIO
- ▸ Amazon ECS Express Mode - Publickey
- ▸ AWS Introduces ECS Express Mode - InfoQ (December 2025)
- ▸ ECS Express Mode discussion - Hacker News